| Abstract [eng] |
The volume and speed of software production is increasing rapidly. So does the complexity of the systems being created. As the number of software systems and their complexity rises, an increase of work required in the field of quality assurance is also observed. One of the core QA activities is testing. Performing testing of complex systems manually is not cost-effective and requires extensive amounts of time. As a result, usage of automated tests becomes more and more widespread. Unfortunately, writing tests also requires a lot of time. Consequently, the need for automatic test generation rises. Nowadays, there is a large amount of test generation algorithms to choose from: symbolic execution, search-based algorithms (e.g. Hill Climbing), genetic algorithms, random generation, model-based generation, etc. The thesis focuses on the area of model-based generation. As a result of this research, an OCL constraints based test generation tool is designed. The tool is implemented as an \"Eclipse\" IDE plug-in, using Java programming language. The tool is then examined to discover what code coverage can the tests it generates achieve on different kinds of software. A set of recommendations for improving the tool, based on the results of aforementioned testing, are provided. After implementing the recommended features, an experiment is carried out to investigate whether the recommendations allowed to improve the tool's performance compared to the initial version. |
